August 30, 2026
Strike selection, timing and volatility all get more attention. Size is the only input that reliably decides whether a run of bad outcomes is an inconvenience or the end of the account.
Premium selling wins most of the time and loses rarely and largely. That distribution does something particular to judgement: the feedback that arrives most often is positive, and it arrives from positions whose size was never tested.
Nine months of every contract expiring worthless is not evidence that the size was right. It's evidence that the size was never the binding constraint, which is a different statement and the one that gets forgotten.
Brokers reduce your available buying power when you open a short option, and the natural reading is that this number represents the risk. It doesn't. It represents the collateral the broker wants against a position, calculated by their formula for their protection, and it's typically far smaller than the actual exposure.
A short $47 put might reduce buying power by a few hundred dollars. The actual obligation is $4,700 of stock. Sizing to the buying power number means sizing to a fraction of the real commitment, and the gap only becomes visible on the day it matters.
Size to the obligation. For a short put that's strike times 100 minus the premium. For a spread it's the width times 100 minus the credit.
The useful sequence runs in the opposite direction from how most positions get opened.
Start: what's the largest position in this single company I'd be content holding through a long drawdown? Say $5,000 on a $60,000 account, about 8%.
At a $47 strike that's one contract, because assignment produces $4,700 of stock. Not two, which produces $9,400 and a 16% single-name concentration acquired on a bad day.
The premium doesn't enter this calculation at all.
Doing it this way means the contract count falls out of a decision you've already made, rather than the premium suggesting a count and the risk being assessed afterwards.
Four puts on four different companies feels like four positions. If they're four regional banks, or four semiconductor names, it's one position with four tickets, and it will behave like one on the day it matters.
The check is to assume that everything you're short in a related group is assigned in the same week, because in a sector repricing that's roughly what happens. Add the obligations. That total is the position size, regardless of how many rows it occupies in the platform.
The same applies to index-correlated positions generally. In a broad selloff, correlations across equities converge toward one, and diversification across names provides much less than it appeared to when the positions were opened.
Size gets adjusted for reasons that feel like analysis and aren't.
After a loss. Increasing size to recover is the reliable predictor of a larger loss, and it's covered at more length in when not to sell premium. The tell is that a number from your own history entered the sizing decision.
After a winning streak. Less discussed and more common. Six consecutive expirations at maximum profit make the previous size feel timid, and the size gets raised on the strength of outcomes produced by the old size.
When premiums are thin. Quiet markets pay less, and the compensating move — more contracts, or closer strikes — delivers the previous income at higher risk. As covered in the piece on high VIX, premium scales with expected movement in both directions, so the low-premium environment isn't underpaying you. It's paying appropriately for a smaller expected move.
Not the only one, but concrete enough to test against your own account: cap the total obligation across all short puts at some fraction of the account, cap any single underlying well below that, and treat correlated names as one underlying for both purposes.
Then write the number down before opening anything, because the number chosen in advance and the number that feels right while looking at an attractive premium are not the same number, and only one of them was chosen with a clear head.